Roughly $1M of development work, open-sourced: Maybe Finance began as a $249/year commercial personal finance product before the company released it all. It aggregates bank accounts, credit cards, loans, investments, crypto, and real estate into a single net worth dashboard with historical trend charts - replacing the spreadsheet that usually glues a whole portfolio together. Transactions are categorized and tagged with rules, with merchant tracking and search across imported or synced activity; budgets track spending by category against plan; and the investment view follows holdings, cost basis, and returns across brokerage accounts. Multi-currency support converts accounts held in different currencies into a single reporting currency, bank synchronization works through Plaid where supported, and manual CSV import covers any institution. An optional AI assistant answers questions grounded in your own financial data. Because the app was built as a paid product with professional design before being open-sourced, its interface quality exceeds most community finance tools - and self-hosting means your balances and transactions are not monetized by a free app or gated behind an annual subscription. The stack is Ruby on Rails with Hotwire on PostgreSQL, licensed AGPL-3.0 and deployed via Docker. The original repository is archived; development continues in the community fork Sure, compatible with the same self-hosted setup.

With over 48,000 GitHub stars and monthly releases since March 2023, LocalAI is the self-hosted AI engine that replaces every OpenAI endpoint with a single Docker container running on your own infrastructure — serving chat completions, image generation, text-to-speech, speech-to-text, embeddings, vision, video generation, and function calling through identical API schemas that require zero application code changes. The composable backend architecture isolates each inference engine as a separate gRPC service running in its own OCI container, so llama.cpp, vLLM, SGLang, transformers, whisper.cpp, diffusers, MLX, Stable Diffusion, and Flux install on demand without touching the core, can run on separate machines, and a fault in one never affects others. Hardware acceleration spans NVIDIA CUDA 12 and 13, AMD ROCm, Intel oneAPI/SYCL, Apple Silicon Metal, Vulkan, and NVIDIA Jetson L4T — or runs entirely on CPU without any GPU. Built-in AI agents support autonomous tool use, retrieval-augmented generation, Model Context Protocol integration, and skill-based workflows directly in the web interface. The model gallery provides curated YAML configuration files for hundreds of models that install with a single command, while P2P federated inference distributes model shards across multiple machines for running models larger than any single node's memory. Multi-user API key authentication with quotas and role-based access enables team deployments. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Grammar, punctuation, and style errors a dictionary lookup can't see: LanguageTool is open-source proofreading powered by a Java rule engine covering English, German, Spanish, French, Portuguese, Dutch, and 25+ other languages. Self-hosting the HTTP server is how you get Grammarly-class checking without sending every sentence you write to a third party - a real concern when the text being proofread is confidential email, legal drafts, or unreleased documentation. Your instance exposes the standard /v2/check API, so the official ecosystem plugs straight in: browser extensions for Chrome and Firefox accept a custom server URL, and integrations exist for VS Code, LibreOffice, Obsidian, Vim, Emacs, and many editors. Notably, self-hosting restores free browser-extension checking that the hosted service moved behind a premium subscription - your server, no character limits, no paywall. Detection quality is tunable: optional n-gram datasets (multi-gigabyte language models for en, de, es, fr, nl) teach the engine word-order and confusion-pair errors like there/their and brakes/breaks, and a fastText model improves automatic language identification. Everything runs offline once models are downloaded. The core is LGPL, the API is documented with Swagger, and rules are community- maintained and constantly expanding.

Named from the German "Ich schrei sonst" - roughly "or I'll scream" - Isso is a lightweight Python/JavaScript commenting server, a drop-in Disqus replacement for people who noticed what Disqus does to reader privacy and page load times. The design premise is printed right in the docs: comments are not Big Data. So the backend is a single SQLite file rather than a database cluster, and the entire client is one embeddable JavaScript file - 65 kB, 20 kB gzipped - that you drop into any static site, blog, or CMS. Commenters write in Markdown, need no account, and can edit or delete their own comments within a configurable window (15 minutes by default). Spam control comes from an optional moderation queue: held comments stay invisible until you activate them via an admin interface or email notification links. Migration is a first-class feature, with importers for Disqus and WordPress exports, so years of existing threads move over intact. Because everything is server-rendered from your own instance, no third party tracks your readers, and real-world switchers report smaller pages and faster loads than the Disqus embed. MIT-licensed, running since 2012.

Perplexity's search experience without Perplexity: Vane deploys Perplexica, an open-source AI answer engine built as the self-hosted alternative. Instead of returning a page of links, it reads your question, searches the live web through the SearxNG metasearch engine, and composes a direct answer with cited sources. Retrieval quality comes from embeddings and similarity search: fetched pages are re-ranked against the query so the model answers from the most relevant passages rather than whatever ranked first. Two query modes cover different needs - Normal mode runs a straightforward web search, while Copilot mode generates multiple reformulated queries and actively pulls content from top matches for harder questions. Focus modes specialize retrieval for academic papers, YouTube, Reddit discussions, Wolfram Alpha calculations, or the general web. The answering model is your choice: OpenAI-compatible APIs or fully local LLMs such as Llama 3 and Mixtral through Ollama, which keeps queries entirely on your infrastructure. Because SearxNG pulls live results, answers reflect current information, and no search history is tracked.

The CAPTCHA bargain - annoy your users and feed their behavior to Google - gets replaced with economics by mCaptcha. Instead of image puzzles, it uses SHA256 proof-of-work: every visitor's browser silently solves a small computational challenge (via a WebAssembly library) before submitting a form. Humans never notice the milliseconds; bots hammering your site must burn more compute sending requests than your server spends answering them, which makes attacks more expensive than defense - the property that also makes mCaptcha genuine DoS protection, not just bot filtering. Written in Rust, the system is fully automated: difficulty scales with traffic, so challenges stay trivial in normal conditions and harden under attack. The privacy and accessibility wins are structural rather than promised: no tracking, no profiling, no user-pattern data collection, and no visual puzzles that exclude users with visual or cognitive impairments - the design was published in Communications of the ACM. Rate limiting is IP-independent, so users behind NATs, VPNs, or Tor get the same experience instead of endless challenge loops, and proofs resist replay attacks, neutering captcha farms. Migration is deliberately easy: the API is compatible with reCAPTCHA and hCaptcha, making it a drop-in replacement. AGPL-licensed core with proprietary-friendly client libraries.

Open the page, write a Markdown note, move on - Memos is a lightweight, self-hosted service built for quick capture. Instead of folders, notebooks, and titles, it presents a timeline: open the page, write a Markdown note, and move on. Notes support headings, code blocks with syntax highlighting, task lists, tables, and file attachments, with tags auto-extracted from #hashtags in the text. Each memo carries a visibility level, private, protected (logged-in users), or public, so one instance works as a personal log, a small team wiki, or a lightweight microblog. The backend is a single Go binary with a React frontend, around 50 MB of memory at runtime and a ~20 MB Docker image, so it fits comfortably on the smallest instance size with near-zero maintenance. SQLite is the default store, with MySQL and PostgreSQL supported for multi-user deployments needing more concurrency, and full REST and gRPC APIs - Connect RPC for browsers, gRPC-Gateway for external tools - make capture scriptable from CLIs, bots, and automation platforms. Fast full-text search spans all memos, pinned notes keep references handy, and a masonry view suits visual browsing. MIT-licensed with zero telemetry; content is stored as plain Markdown in a database you control, so notes remain readable, exportable, and free of proprietary formats.

Over 400 technologies and thousands of smart devices from any manufacturer, unified under one roof: openHAB is the vendor-neutral home automation platform with a pluggable binding architecture. Each binding translates a device or service into openHAB's clean abstraction: Things expose Channels, Channels link to Items, and Items feed a rules engine that runs your home. That engine meets you at your skill level: Blockly gives non-programmers drag-and-drop visual logic, JS Scripting (GraalJS with the openhab-js library) is the modern text-based standard, the classic Rules DSL remains supported, and JSR223 opens the door to Python, Ruby, and Groovy. Time- and event-based triggers, scripts, notifications, and voice control compose into automations of any complexity, and users report decade-old rule sets still running rock solid. The Main UI handles configuration, semantic modeling, and now built-in charting - no external Grafana required. Built in Java on Apache Karaf's OSGi runtime and stewarded by the non-profit openHAB Foundation, it requires no cloud to function: everything runs locally, talking directly to your devices. Optional connectors bridge to Alexa, Google Assistant, and HomeKit, with iOS, Android, and web apps for control from anywhere.

The utilities engineers otherwise scatter across a dozen ad-laden websites - 80+ of them - live together in IT-Tools, one fast, polished web app. Crypto covers JWT decoding, MD5 through SHA-512 hashing, HMAC and bcrypt generation, RSA key pairs, and password strength analysis. Converters handle JSON to CSV, YAML, and TOML, Base64 files, URL encoding, HTML entities, color formats, and Docker run commands to Compose files. Generators produce UUIDv4, ULID, BIP39 mnemonics, QR codes (including Wi-Fi QR), and tokens; text tools include a regex tester, diff viewer, slug and case converters; web utilities parse URLs and user agents, look up HTTP status codes and MIME types, and inspect Open Graph metadata; plus a cron parser, chmod calculator, and more. The privacy argument is the point: JWTs contain user IDs, hashes derive from passwords, JSON dumps hold PII - exactly the inputs you least want a third-party utility site to log. IT-Tools is a frontend-only static bundle (Vue/TypeScript, GPL-3.0, 39k+ GitHub stars) served by Nginx in one container, so everything runs client-side on your infrastructure with nothing transmitted anywhere. New tools ship roughly monthly, and a scaffolding script makes adding custom ones straightforward.

With over 110,000 GitHub stars and one of the fastest-growing open-source communities in the self-hosted space, Immich delivers a Google Photos-grade experience entirely on your own hardware. The platform handles automatic background backup from Android and iOS devices, deduplication, and support for RAW formats, LivePhotos, and MotionPhotos. Its machine learning pipeline runs facial recognition and clustering locally on your server, enabling you to group photos by person without sending a single image to the cloud. CLIP-based semantic search lets you find images by describing their content in natural language, while metadata-driven search covers EXIF data, dates, and locations. The web interface built with SvelteKit provides a responsive timeline view, albums, shared albums with configurable permissions, public sharing links with optional passwords and expiry dates, partner sharing for family libraries, and a global map plotting photos by GPS coordinates. Administrative features include multi-user support with per-user storage quotas, OAuth integration, API key management, and a user-defined storage structure for organizing files on disk. The architecture uses PostgreSQL for metadata, Redis with BullMQ for background job queues handling thumbnail generation, video transcoding, and smart search indexing, and exposes over 400 REST API endpoints documented via OpenAPI with auto-generated SDKs for web, mobile, and CLI clients. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. AGPL-3.0 licensed.

The Bitwarden server, reimplemented in Rust: Vaultwarden (formerly bitwarden_rs) is the unofficial lightweight edition. It speaks the same wire protocol as the official server, so every official Bitwarden client - browser extensions, iOS, Android, desktop, and the bw CLI - connects without modification, while the server itself runs as a single container against SQLite (or MySQL/MariaDB/PostgreSQL) instead of the official multi-container stack that wants gigabytes of RAM. Features Bitwarden gates behind paid tiers ship free: organizations with collections, groups, member roles, and policies; TOTP code storage; file attachments; Bitwarden Send; Emergency Access; event logs; and admin password reset. Two-factor options cover authenticator apps, email, FIDO2 WebAuthn, YubiKey, and Duo, and OIDC-based SSO landed natively in v1.35.0. Zero-knowledge encryption is unchanged - vault data is encrypted client-side and the master password never reaches the server. Attachments and Sends store on local disk or S3-compatible backends, an admin panel manages users and server settings, and backup is copying one data directory. Suited to individuals and teams up to roughly 50 users.

Your KeePass vaults, opened from any browser: KeeWeb reads, edits, and creates standard KDBX files, so it works with the same databases as KeePass and KeePassXC without conversion or lock-in. Self-hosting the web app gives you a password manager reachable from any modern browser, including mobile, with no client installation and no third-party cloud in the loop. All KDBX cryptography runs client-side; the server just serves the static app. Open multiple vault files simultaneously and search them all from one box, with advanced options covering specific fields, password history, and regular expressions. Vaults load from local files, your own server (WebDAV), or Dropbox, Google Drive, and OneDrive, with automatic sync - and files are cached for offline use, so a dropped connection never locks you out; changes resync once you're back online. Day-to-day niceties include a configurable password generator, protected fields that stay masked and are held in memory more defensively, entry history, tags with easy input, drag-and-drop attachments, and per-entry icons with favicon fetching. The optional KeeWeb Connect extension (Chrome, Firefox, Edge, Safari) autofills credentials using the keepassxc-protocol. MIT-licensed with matching desktop apps for macOS, Windows, and Linux.

Live web search plus an LLM of your choice: Farfalle is an open-source, self-hosted answer engine in the Perplexity mold. Queries route through one of several search providers - self-hosted SearXNG for a fully independent stack, or Tavily, Serper, and Bing APIs - and the model composes a cited answer from the retrieved results. Model flexibility is the core design: run llama3, mistral, gemma, or phi3 locally through Ollama for zero per-query cost and full privacy, use cloud models like GPT-4o or Groq-hosted Llama 3 for speed, or route to any provider via LiteLLM. An Expert Search mode uses an agent that plans a multi-step search strategy and executes it for harder questions, and chat history keeps prior research sessions available. The stack is a Next.js and shadcn/ui frontend over a FastAPI backend with Redis rate limiting, shipped as a pre-built Docker image. A browser search-engine entry pointing at your instance makes it the default search from the address bar. Paired with SearXNG and Ollama, the whole pipeline runs with no external API at all.

Merge, split, compress, convert, edit, annotate, redact, OCR, and sign PDFs - BentoPDF packs over 130 tools into a privacy-first toolkit that runs entirely in the browser through WebAssembly. Files are never uploaded - processing happens in browser memory on the user's machine and disappears when the tab closes, which makes the tool GDPR-clean by architecture and safe for financial, legal, and internal documents. The engine combines WASM builds of PyMuPDF, Ghostscript, and CoherentPDF; Tesseract handles OCR with searchable text-layer output; Office conversions cover Word, Excel, and PowerPoint; and digital signatures use X.509 certificates (PFX/PEM) with the private key staying on the client. Because there is no server-side processing, deployment is a static-file exercise: a single Docker container, or any static host. A dedicated self-hosted build strips the marketing pages while keeping every tool, and air-gapped deployments are first-class - an automated script bundles the WASM modules, OCR language data, and fonts for fully offline networks. No accounts, no limits, no watermarks; TypeScript and Vite under the hood.

With over 10,800 GitHub stars earned in under two years, Blinko has emerged as the privacy-first answer to cloud-dependent note-taking tools by combining instant thought capture with AI-powered retrieval that actually understands what you wrote rather than just matching keywords. The core engine uses Retrieval-Augmented Generation to build vector embeddings of every note in your PostgreSQL database, enabling natural language queries like "what were my thoughts on the database migration last Tuesday" to surface relevant content through semantic understanding rather than exact string matching. AI integration supports both cloud providers — OpenAI, MiniMax, and compatible endpoints — and fully local inference through Ollama running models like Llama 3.2 on your own hardware, ensuring your notes never leave your network when privacy demands it. Built on Next.js with a React frontend, the web interface presents notes as cards with full Markdown support including code blocks, LaTeX, and rich formatting, with a clean input bar for capturing fleeting thoughts in seconds. The Tauri-based desktop and mobile clients extend access to macOS, Windows, Linux, and Android with native performance characteristics. Notes are stored as plain text in PostgreSQL with vector indexes that update incrementally as new content arrives, and the embedding index can be rebuilt on demand when switching between AI providers. Multiple users can share a single instance with individual accounts, and data exports to standard formats. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GPL-3.0 licensed.

With over 38,000 GitHub stars, nearly one million registered users, and over 1.3 million resumes created, Reactive Resume delivers a professional resume creation experience that rivals commercial platforms like Resume.io and Zety without collecting user data, showing advertisements, or imposing paywall restrictions. The real-time WYSIWYG editor renders changes instantly as users type, providing immediate visual feedback across all resume sections including work experience, education, skills, certifications, projects, and custom sections with typed extensions. Over fifteen professionally designed templates with full color customization, font selection, spacing adjustment, and icon style selectors ensure each resume reflects individual style preferences. The drag-and-drop system lets users move items between sections and pages freely across multi-column and full-width layouts. PDF generation produces pixel-perfect documents using a headless Chromium rendering pipeline, ensuring consistent output across all devices and print settings. Multi-language support covers over twenty languages with RTL script handling for Arabic and Hebrew. Authentication supports passkeys and multiple OAuth providers via Better Auth, while self-hosted deployment runs via Docker Compose with PostgreSQL and Browserless for PDF rendering. The import system accepts JSON Resume format, LinkedIn data exports, and PDF/DOCX files when AI integration is configured. AI-powered content suggestions using OpenAI, Google Gemini, Anthropic Claude, or local Ollama models help improve bullet points and professional summaries. A CSS editor with autocompletion allows fine-grained styling control. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

No cookies, no fingerprinting, no cross-site tracking, no personal data collection - Umami's privacy contract is the foundation of the open-source web analytics platform. IP addresses are hashed rather than stored, which makes it GDPR, CCPA, and PECR compliant by default - the consent banner can come off the site entirely. The tracking script is under 2 KB, roughly 20x smaller than Google Analytics, so measurement stops being a page-weight tax. The dashboard covers the core metrics - pageviews, visitors, bounce rate, visit duration, referrers, browsers, devices, and countries - with any date range and filtering by country or device. Beyond pageviews, custom events track clicks, form submissions, and signups via a data attribute or one JavaScript call, and advanced reports add funnels, user journeys, retention and cohort analysis, goals, and automatic UTM campaign tracking. Anonymous session views show individual visitor activity without identifying anyone. Teams share websites with role-based access, one instance manages unlimited sites, and a full REST API exposes every metric programmatically. MIT-licensed and self-hosted on PostgreSQL or MySQL via Docker, your analytics data never leaves your infrastructure.

Built as a direct rejection of the adtech model, Plausible is the best-known privacy-first web analytics tool - lightweight, cookie-free, and open-source. It sets no cookies and stores no personal data: unique visitors are counted via a hash of IP plus User-Agent that rotates every 24 hours and is never stored raw, so no consent banner is required and GDPR compliance is structural rather than contractual. The tracking script is under 1 KB - orders of magnitude lighter than GA - and the dashboard is a deliberate contrast to GA4's sprawl: one fast-loading page with visitors, sources, top pages, countries, devices, and UTM breakdowns, filterable by any dimension. Custom events and goals track signups and clicks, Google Search Console integration pulls in search queries, scheduled email reports keep stakeholders updated, and the Stats API (v2) plus CSV export feed data anywhere. This is the AGPL-licensed Community Edition, the same Elixir codebase that powers Plausible's cloud service, running as three containers: the web app, PostgreSQL for accounts, and ClickHouse for event storage - which means self-hosters get direct SQL access to raw analytics data the cloud version never exposes. Traffic data stays entirely on your server, with no visitor caps or per-pageview pricing.